Emotion Recognition for Low-Resource Turkish: Fine-Tuning BERTurk on TREMO and Testing on Xenophobic Political Discourse

TL;DR

This work develops a Turkish Emotion Recognition Model (ERM) by fine-tuning BERTurk on the TREMO dataset to classify six basic emotions (happiness, fear, anger, sadness, disgust, surprise). It collects 47,024 Turkish tweets about Sessiz Istila from X (June 2021–Dec 2022), applies careful cleaning and emoji normalization, and balances TREMO to enable robust training. The model achieves $92.62\%$ accuracy on a held-out test and reveals strong domain insights: anger and surprise dominate Turkish discourse on refugees, with a temporal shift from surprise (2021) to anger (2022). This approach advances Turkish NLP in low-resource settings and demonstrates practical utility for real-time sentiment tracking across marketing, PR, and crisis management, while outlining methodological avenues to broaden emotion taxonomies and contextual modeling.

Abstract

Social media platforms like X (formerly Twitter) play a crucial role in shaping public discourse and societal norms. This study examines the term Sessiz Istila (Silent Invasion) on Turkish social media, highlighting the rise of anti-refugee sentiment amidst the Syrian refugee influx. Using BERTurk and the TREMO dataset, we developed an advanced Emotion Recognition Model (ERM) tailored for Turkish, achieving 92.62% accuracy in categorizing emotions such as happiness, fear, anger, sadness, disgust, and surprise. By applying this model to large-scale X data, the study uncovers emotional nuances in Turkish discourse, contributing to computational social science by advancing sentiment analysis in underrepresented languages and enhancing our understanding of global digital discourse and the unique linguistic challenges of Turkish. The findings underscore the transformative potential of localized NLP tools, with our ERM model offering practical applications for real-time sentiment analysis in Turkish-language contexts. By addressing critical areas, including marketing, public relations, and crisis management, these models facilitate improved decision-making through timely and accurate sentiment tracking. This highlights the significance of advancing research that accounts for regional and linguistic nuances.

Figures (11)

• Figure 1: Research flow.
• Figure 2: Initial emotion dataset in TREMO.
• Figure 3: Calibrated dataset in TREMO.
• Figure 4: Training and validation loss convergence.
• Figure 5: Model convergence in validation accuracy and precision.